diff options
author | Wael El-Essawy <welessa@us.ibm.com> | 2016-10-27 12:54:20 -0500 |
---|---|---|
committer | Wael El-Essawy <welessa@us.ibm.com> | 2016-11-03 15:08:04 -0400 |
commit | 45ceb3d13361ac099c7b0b9f2ff51a731e296ed1 (patch) | |
tree | 0b9edc946040fe0f3bd2b2f82a09b1f3d94ebc56 /src/occ_405/amec/amec_init.c | |
parent | aa7187f24959d2ec1fbee81a5d100a749161096f (diff) | |
download | talos-occ-45ceb3d13361ac099c7b0b9f2ff51a731e296ed1.tar.gz talos-occ-45ceb3d13361ac099c7b0b9f2ff51a731e296ed1.zip |
Implement task_poke_watchdogs
task_poke_watchdogs() should be called every 2ms (ticks 1 and 9) on both
master and slaves while in observation and active state and do the following:
1. Every time called: Enable/Reset the OCC heartbeat: done by a write to
OCB OCC Heartbeat Register (set count to 8ms)
2. Every time called: Reset memory deadman timer for 1 MCA (skip if not
present and just wait until next call to check next MCA to keep same
timing of reset per MCA regardless of # present) Resetting the deadman
is done by reading one of the memory performance counters, use one at
SCOM offset 0x13C. NOTE: Will take 16ms (8 MCAs x 2ms) to reset all
memory timers, this is fine since the shortest time the deadman timeout
can be configured to is 28ms
3. Every 4ms (on tick 1 only) : Verify PGPE is still functional by reading
PGPE Beacon from SRAM if after 8ms (2 consecutive checks) there is no
change to the PGPE Beacon count then log an error and request reset.
In addition, this commit adds entries for the PGPE image header and shared
SRAM in the TLB, and partially reads PGPE image header parameters.
Change-Id: I9906102b3349506612d55c57e9f5c28441eaeb39
RTC: 154960
Reviewed-on: http://ralgit01.raleigh.ibm.com/gerrit1/31916
Tested-by: FSP CI Jenkins <fsp-CI-jenkins+hostboot@us.ibm.com>
Reviewed-by: Wael El-Essawy <welessa@us.ibm.com>
Diffstat (limited to 'src/occ_405/amec/amec_init.c')
-rw-r--r-- | src/occ_405/amec/amec_init.c | 10 |
1 files changed, 5 insertions, 5 deletions
diff --git a/src/occ_405/amec/amec_init.c b/src/occ_405/amec/amec_init.c index 765b005..b338915 100644 --- a/src/occ_405/amec/amec_init.c +++ b/src/occ_405/amec/amec_init.c @@ -393,12 +393,12 @@ void amec_slave_init() (void *) GPE_ENGINE_1, //callback argument ASYNC_CALLBACK_IMMEDIATE ); //options - // If we couldn't create the poreFlex objects, there must be a major problem + // If we couldn't create the GpeRequest objects, there must be a major problem // so we will log an error and halt OCC. if( rc || rc2 ) { - //If fail to create pore flex object then there is a problem. - TRAC_ERR("Failed to create GPE duration poreFlex object[0x%x, 0x%x]", rc, rc2 ); + //If fail to create GpeRequest object then there is a problem. + TRAC_ERR("Failed to create GPE duration GpeRequest object[0x%x, 0x%x]", rc, rc2 ); /* @ * @errortype @@ -407,7 +407,7 @@ void amec_slave_init() * @userdata1 return code - gpe0 * @userdata2 return code - gpe1 * @userdata4 OCC_NO_EXTENDED_RC - * @devdesc Failure to create PORE-GPE poreFlex object for FW timing + * @devdesc Failure to create GpeRequest object for FW timing * analysis. * */ @@ -416,7 +416,7 @@ void amec_slave_init() SSX_GENERIC_FAILURE, //reasoncode OCC_NO_EXTENDED_RC, //Extended reason code ERRL_SEV_PREDICTIVE, //Severity - NULL, //TODO: create trace //Trace Buf + NULL, //Trace Buf DEFAULT_TRACE_SIZE, //Trace Size rc, //userdata1 rc2 //userdata2 |